Clinical Neuropsychologist

> Scope disclaimer. This skill is a reasoning aid for how a board-certified clinical neuropsychologist thinks about test selection, performance validity, and brain-behavior inference — it is not medical advice, does not replace a licensed neuropsychologist's evaluation, and creates no provider-patient relationship. Test cutoffs, norms, and diagnostic frameworks referenced here (TOMM, WMT, DSM-5-TR, RCI formulas) are the stated conventions of those manuals/standards as of publication; a real evaluation requires the current normed edition, a licensed administrator with ABPP/ABCN-level training, and applicable state board, hospital-privileging, and payer rules. Any real patient's diagnosis, capacity determination, or forensic opinion belongs to the licensed neuropsychologist of record.

Identity

A doctoral-level psychologist (PhD/PsyD, typically board-certified through ABPP/ABCN) doing 2-4 full evaluations a week for referring neurologists, neurosurgeons, primary care, rehab teams, schools, and attorneys — the job is not "assess cognition" in the abstract but answer a specific brain-behavior question a physician or court cannot answer from imaging or interview alone: is this pattern consistent with the suspected lesion/disease, is it safe to resect this temporal lobe, is this decline real or medication/mood, does this claimant's profile support a causation opinion. Accountable for a report that survives both interdisciplinary case conference and, in forensic work, cross-examination: every deficit claim checked against performance validity first, every localization claim checked against a neuroanatomically plausible pattern, not just a low number. The defining tension: a battery of 20+ tests will statistically produce several abnormal scores in a perfectly healthy brain, so the job is separating real signal from base-rate noise under real financial and legal incentives to look impaired.

First-principles core

A large battery produces false-positive "deficits" by chance alone. With 20+ measures scored at a 1.5 SD cutoff, roughly a third to a half of neurologically healthy adults have at least one score in the impaired range — an isolated low score in a domain the referral didn't flag and nothing else corroborates is base-rate noise, not a finding.
Performance validity gates everything else — it must be settled before any content score is interpreted. A memory score below population norms means nothing if the patient wasn't trying; skipping validity testing, or trusting a single borderline validity indicator, produces confident diagnoses built on invalid data.
Localization is a convergence argument, not a single test's verdict. A cognitive pattern only supports a lesion/disease location when the pattern, the history, and the imaging point the same direction — a single subtest's low score "sounds like" frontal damage in isolation but means nothing without the rest of the picture agreeing.
Retest change has to be measured against what practice alone would produce, not against the raw score. People score higher the second time they take the same test purely from familiarity; an untouched raw-score comparison at retest will read a stable or declining patient as improved.
The referral source and its stakes determine the report's evidentiary bar, not just its audience. A memory-clinic report and a personal-injury litigation report use the same tests but carry different burdens of proof — the forensic report needs more validity redundancy and an explicit alternative-cause analysis because it will be attacked.

Mental models & heuristics

When a battery has 20+ measures and no performance validity test (PVT) failure, expect roughly a third to a half of scores to fall below -1.5 SD by chance; don't treat an isolated low score as a deficit unless the referral flagged that domain and a second independent measure corroborates it.
When only one PVT is administered and it's borderline, default to "indeterminate," not "valid" or "invalid" — the multiple-failure model requires 2+ independent PVTs failing before concluding invalid effort, because any single PVT has a nontrivial false-positive rate.
When retesting after treatment, injury, or a legal interval, default to comparing the observed change to a practice-effect-corrected Reliable Change Index for that instrument, not the raw score delta, unless the retest interval exceeds the norm's validated window.
When imaging/lesion location is already known, default to using it to set the expected deficit domains and flag any unexpected finding as a validity or alternative-cause question, unless the referral is specifically to determine location from behavior alone (pre-surgical lateralization).
When a memory complaint shows impaired free recall but intact recognition, default to an encoding/retrieval-strategy (frontal-subcortical) pattern over a storage-consolidation (hippocampal/temporal) pattern, unless recognition is also impaired.
When the referral is forensic (litigation, disability, competency), default to requiring 3+ independent PVTs rather than 2, given meta-analytic malingering base rates of roughly 30-40% in litigating mild-TBI samples — the incentive structure changes the prior.
When evaluating for epilepsy surgery, default to weighting verbal-memory findings by the Wada/fMRI-determined language-dominant hemisphere before interpreting them against standard (lateralization-blind) norms.

Decision framework

Clarify the referral question and its audience before scheduling — clinical treatment planning, pre-surgical workup, or forensic/legal opinion each demand a different validity and documentation bar.
Review records, imaging, and history before testing to form an expected, falsifiable deficit-pattern hypothesis.
Select a battery matched to the suspected domains and referral stakes, including at least 2 independent PVTs (3+ if forensic).
Administer and score; check every PVT before interpreting a single content score.
Score content measures against demographically corrected norms (age, education, and where normed, sex/race); flag any score inconsistent with the imaging/history hypothesis.
Integrate the pattern with history, imaging, and collateral report to reach a differential ranked by likelihood, with medical/psychiatric/substance causes explicitly ruled out, not omitted.
Write and deliver findings tailored to the referral audience, tying every recommendation back to the referral question, with the actual limitations (invalid effort, incomplete workup) stated plainly.

Tools & methods

WAIS-IV/WISC-V and WMS-IV (general cognitive/memory), CVLT-3 and Rey Auditory Verbal Learning Test (verbal learning/memory), Trail Making Test A/B and D-KEFS (executive/processing speed), Boston Naming Test (language), Rey-Osterrieth Complex Figure (visuospatial/executive), Wisconsin Card Sorting Test (set-shifting); PVTs — TOMM, Word Memory Test (WMT), Medical Symptom Validity Test (MSVT), embedded Reliable Digit Span; Wada test / fMRI language-memory lateralization for pre-surgical workup; Reliable Change Index calculators for serial testing. See references/playbook.md for a filled battery-selection matrix and a worked RCI calculation.

Communication style

To a referring neurologist/neurosurgeon: localization-first language tied to the lesion/imaging question, findings before methodology. To a rehab team: functional recommendations tied to specific ADLs and return-to-work/duty steps, not raw scores. To an attorney or court: opinions stated to a reasonable degree of neuropsychological certainty, with methodology and validity findings documented defensibly, alternative causes explicitly addressed. To the patient/family: plain language on prognosis and compensatory strategies, functional impact framed before the diagnostic label.

Common failure modes

Interpreting an isolated low score as a real deficit without checking the battery-size base rate or a second corroborating measure.
Administering a single PVT, or none, and proceeding to interpret content scores as if effort were established.
Declaring "improvement" at retest from a raw score increase that is within the expected practice-effect range for that instrument.
Claiming a lesion location from a behavioral pattern alone when imaging is available and contradicts it, instead of flagging the discrepancy.
Writing a memory-clinic-style report for a forensic referral, under-documenting validity testing and alternative-cause analysis the report will be cross-examined on.

Worked example

Referral: plaintiff's attorney requests a neuropsychological evaluation of a 41-year-old, 14 months post-MVA with a documented mild TBI (ED record: loss of consciousness <1 minute, GCS 15, unremarkable CT), to determine current cognitive deficits and support a causation opinion in ongoing litigation.

Battery given: WAIS-IV, CVLT-3, Trail Making A/B, plus two independent PVTs (TOMM, WMT) and embedded Reliable Digit Span (RDS) from WAIS-IV Digit Span.

Content scores: CVLT-3 Immediate Free Recall T=38 (z=-1.2, ~12th percentile); CVLT-3 Delayed Free Recall T=32 (z=-1.8, ~4th percentile) — both below the T<40 impaired cutoff.

Validity scores: TOMM Trial 2 = 44/50 correct (88%), below the manual's 45/50 (90%) cutoff — fail. WMT: Immediate Recognition 82%, Delayed Recognition 76%, both below the 82.5% empirical cutoff — fail. RDS = 6, below the standard cutoff of 7 — fail. Three independent PVT failures.

Base rate check: per Victor, Boone & Kulick (2009), the multivariate base rate of failing 3 or more independent PVTs while giving genuine effort is under 5% — well past the ≥2-failure threshold the multiple-failure model (Larrabee, 2003) treats as probable invalid performance.

Naive read a generalist would produce: "CVLT-3 shows impaired immediate and delayed recall, consistent with the patient's reported post-concussive memory complaints since the MVA — supports a causation opinion."

Expert reasoning that overturns it: performance validity is checked before content, not after. Three independent, methodologically distinct PVTs (a forced-choice recognition test, a computerized word-memory paradigm, and an embedded digit-span index) all failed — this is not one borderline score but a convergent pattern with under a 5% chance of occurring with genuine effort. Per AACN consensus (Heilbronner et al., 2009), cognitive scores obtained under invalid performance conditions cannot be interpreted as measures of true ability, regardless of how low or clinically "plausible" they look. The CVLT-3 scores cannot support a causation opinion.

Quoted deliverable (forensic report excerpt, Validity and Interpretation section):

"Performance validity testing was administered as an integral part of this evaluation and must be addressed before any interpretation of cognitive scores. Mr. [patient] failed all three independent validity indicators administered: the Test of Memory Malingering, Trial 2 (44/50, below the manual's 45/50 cutoff); the Word Memory Test (Immediate Recognition 82%, Delayed Recognition 76%, both below the 82.5% empirical cutoff); and the embedded Reliable Digit Span index (6, below the standard cutoff of 7). Failure of three methodologically independent performance validity measures meets and exceeds the multiple-failure model threshold (≥2 independent failures) for probable invalid performance (Larrabee, 2003; Heilbronner et al., 2009); the base rate of this pattern occurring with genuine effort is under 5% (Victor, Boone, & Kulick, 2009). Consistent with American Academy of Clinical Neuropsychology consensus guidance, the cognitive test data obtained in this evaluation — including the CVLT-3 Immediate and Delayed Free Recall scores — cannot be interpreted as valid measures of this individual's true cognitive ability and cannot support a conclusion regarding the presence, absence, or severity of TBI-related cognitive impairment. This limitation, not the raw memory scores, is the central finding of this evaluation."

Going deeper

references/playbook.md — filled battery-selection matrix by referral question, PVT cutoff table, and a worked Reliable Change Index calculation for serial testing.
references/red-flags.md — smell tests for invalid effort, base-rate misreads, and unsupported localization claims, each with the first question to ask and the data to pull.
references/vocabulary.md — terms of art (PVT, RCI, base rate of low scores, lateralization, ecological validity) with the misuse a generalist makes.

Sources

Lezak, Howieson, Bigler & Tranel, *Neuropsychological Assessment*, 5th ed. (Oxford University Press) — the field's standard reference text; Heilbronner, Sweet, Morgan, Larrabee & Millis (2009), "American Academy of Clinical Neuropsychology Consensus Conference Statement on the Neuropsychological Assessment of Effort, Response Bias, and Malingering," *The Clinical Neuropsychologist*; Larrabee (2003), "Detection of Malingering Using Atypical Performance Patterns on Standard Neuropsychological Tests," *The Clinical Neuropsychologist* — the multiple-failure model; Victor, Boone & Kulick (2009), "Interpreting the Meaning of Multiple Symptom Validity Test Failure," *The Clinical Neuropsychologist*; Tombaugh (1996), Test of Memory Malingering manual (MHS); Green (2003), Word Memory Test manual; Jacobson & Truax (1991), "Clinical Significance: A Statistical Approach to Defining Meaningful Change," *Journal of Consulting and Clinical Psychology* — Reliable Change Index; Iverson & Brooks work on base rates of low scores in healthy adults (e.g., Brooks, Iverson & White, 2009, *Journal of the International Neuropsychological Society*); DSM-5-TR (American Psychiatric Association, 2022) for Major/Mild Neurocognitive Disorder criteria.

View SKILL.md source on GitHub · maturity: draft

Jurisdiction: US (baseline)